Lake SST with ECMWF HRES

[dynames]

Hi there,

I am trying to do 1-day simulations using ECMWF HRES 0.1x0.1 hourly lat/lon grid data and analyze the SST over a lake on the center of my domain.
I'm using Analysis product for the 00 UTC data and Forecast product for the other 23 steps. WRF Version is 4.1.5.
When I run metgrid.exe with default METGRID.TBL and plot SST field of met_em files, while between 02 and 24 UTC the SST field over lake represents the LANDMASK, it is different than LANDMASK and coarse on 00 and 01 UTC met_em files. I've done a research and found that changing METGRID.TBL might work. So I made these modifications to my METGRID.TBL:

DEFAULT METGRID.TBL
---------------------------
name=SST
        interp_option=sixteen_pt+four_pt
        fill_missing=0.
        missing_value=-1.E30
        flag_in_output=FLAG_SST

MODIFIED METGRID.TBL
----------------------------
name=SST 
	interp_option=sixteen_pt+four_pt+wt_average_4pt+wt_average_16pt+search
	missing_value=-1.E30
	fill_missing=0.
	flag_in_output=FLAG_SST
	masked=land
interp_mask=LANDMASK(1)

After this modification I get met_em files that represents the LANDMASK well on all time steps. So I move on to real.exe and wrf.exe. I'm using sst_update=1.
Simulation is completed successfully. However, after I plot SST field of wrfout files, SST pattern of 00 UTC and 01 UTC is completely different from the SST pattern of other timesteps. SST of 0 and 1 UTC seems reasonable however on other timesteps, there are very cold temperature values on the outer boundary of the lake. Thus, that low temperature values are interpolated to other small water areas that are close to the lake, so those water points have wrong values as well.
What might be the problem?
Im attaching wrfout output of 00 UTC (0.png) and 02 UTC (2.png) in order for you to see the difference in the patterns. "0utcdef.png " and "2utcdef.png" are example of wrfout with default METGRID.TBL. I'm attaching my namelist.wps and namelist.input file as well.

Thanks in advance.

 

[Ming Chen]

Would you please look SST over lake at your met_em files for all the 24 hours? Since you run the case with sst_update = 1, I suppose that SST from met-em files will be ingested into WRF. If the lake SST is cold in met_em files, then they will be cold in wrfout, too.
Please keep me updated about the case. Thanks.

 

[dynames]

Would you please look SST over lake at your met_em files for all the 24 hours? Since you run the case with sst_update = 1, I suppose that SST from met-em files will be ingested into WRF. If the lake SST is cold in met_em files, then they will be cold in wrfout, too.
Please keep me updated about the case. Thanks.

I've plotted met_em SST. I'm attaching them. I've also changed interp_mask = LANDMASK(1) to LANDSEA(1). Thats true right? This way, interpolation will be from the water points of ECMWF data instead of the points from geogrid's landmask as far as I understand. I am attaching result of both of them (landmask=1 and landsea=1). Result of "LANDSEA=1" seems better than the other. Also the default interp_option which is:

"interp_option=sixteen_pt+four_pt"

is definitely not working for my case so I use:

"interp_option=sixteen_pt+four_pt+wt_average_4pt+wt_average_16pt+search"

Is there a better option?
On the landsea=1 plot, white areas are higher than 9 °C which are reasonable, however there are some 3 °C values on the lower left corner of the lake. I checked the ecmwf data to compare (ecmwf_sst.png) and it seems there are low values over there so it matches. However the water area on the bottom left of the plot is completely cold, I won't do calculations with that water area however that concerns me a little bit. Is it normal for that water area to be cold?

Thanks in advance.

 

[Ming Chen]

Hi,
Thanks for the detailed description. I apologize for answering late.
The results look much better when using the LANDSEA(1) as mask. I believe this is the correct way to handle the inconsistency between WPS landmask and lands from ECMWF.
For the question related to the cold water area, unfortunately I don't have an answer without looking into details of the case and data. I am thinking it is better to compare with original ECMWF. Then we may get some idea whether it is fictitious or it is reasonable.